Neuropsychologia 51 (2013) 2781–2790

Contents lists available at ScienceDirect

Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia

Increased parietal activity after training of interference control
Stephan Oelhafen a,n, Aki Nikolaidis a,b, Tullia Padovani a, Daniela Blaser a,
Thomas Koenig c, Walter J. Perrig a,d
a

Institute of Psychology, University of Bern, Switzerland
Neuroscience Program, University of Illinois Urbana Champaign, United States
c
Department of Psychiatric Neurophysiology, University Hospital of Psychiatry, University of Bern, Switzerland
d
Center for Cognition, Learning and Memory, University of Bern, Switzerland
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 21 August 2012
Received in revised form
13 August 2013
Accepted 16 August 2013
Available online 25 August 2013

Recent studies suggest that computerized cognitive training leads to improved performance in related
but untrained tasks (i.e. transfer effects). However, most study designs prevent disentangling which of
the task components are necessary for transfer. In the current study, we examined whether training on
two variants of the adaptive dual n-back task would affect untrained task performance and the
corresponding electrophysiological event-related potentials (ERPs). Forty three healthy young adults
were trained for three weeks with a high or low interference training variant of the dual n-back task, or
they were assigned to a passive control group. While n-back training with high interference led to partial
improvements in the Attention Network Test (ANT), we did not find transfer to measures of working
memory and fluid intelligence. ERP analysis in the n-back task and the ANT indicated overlapping
processes in the P3 time range. Moreover, in the ANT, we detected increased parietal activity for the
interference training group alone. In contrast, we did not find electrophysiological differences between
the low interference training and the control group. These findings suggest that training on an
interference control task leads to higher electrophysiological activity in the parietal cortex, which may
be related to improvements in processing speed, attentional control, or both.
& 2013 Elsevier Ltd. All rights reserved.

Keywords:
Cognitive training
Interference control
Electrical neuroimaging
Event-related potential (ERP)
Parietal cortex
n-back

1. Introduction
A fundamental question in the study of learning is whether
acquired knowledge or skills transfer to new but similar tasks or
situations. Formerly, many researchers agreed that in general, little
transfer occurs (Detterman, 1993); recently, the transfer of learning has received more scientific attention due to encouraging
approaches to train core mechanisms of working memory (WM).
These forms of WM training have improved performance in
related but untrained tasks, such as measures of fluid intelligence
(gF; Jaeggi, Buschkuehl, Jonides, & Perrig, 2008; Jaeggi,
Buschkuehl, Jonides, & Shah, 2011; Klingberg et al., 2005). However, inconsistencies in methodology and results have raised
questions about the efficacy of WM interventions for the enhancement of fluid intelligence (cf. Morrison & Chein, 2011; Redick et al.,
2012). While the factors promoting transfer are still unknown,
many have argued that transfer of learning is possible to the
degree that the training and transfer tasks involve overlapping
neural networks or share cognitive mechanisms (Dahlin,
n
Correspondence to: University of Bern, Department of Psychology, Division of
Experimental Psychology and Neuropsychology, Fabrikstrasse 8, 3012 Bern,
Switzerland. Tel.: þ 41 316314734.
E-mail address: stephan.oelhafen@psy.unibe.ch (S. Oelhafen).

0028-3932/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.neuropsychologia.2013.08.012

Stigsdotter Neely, Larsson, Bäckman, & Nyberg, 2008; Jonides,
2004; Klingberg, 2010; Perrig, Hollenstein, & Oelhafen, 2009;
Persson & Reuter-Lorenz, 2008).
Methodological issues in several WM training studies have
presented challenges to the replication and generalization of their
findings (cf. Shipstead, Redick, & Engle, 2010). First, since cognitive
training studies typically involve complex training paradigms or
batteries of training tasks, isolating the cognitive processes critical
for transfer proves difficult (Jaeggi et al., 2008; Klingberg et al.,
2005; Olesen, Westerberg, & Klingberg, 2004; Thorell, Lindqvist,
Bergman, Bohlin, & Klingberg, 2009). A second, related issue
concerns the adaptive adjustment of difficulty in many training
regimens. Since these are often compared with non-adaptive
training tasks, the intended differences may be confounded with
decreased motivation and training compliance in active control
groups. These unintended differences between training and active
control groups impair internal validity (cf. Shipstead et al., 2010;
Shipstead, Redick, & Engle, 2012). Therefore, contrasting the
training intervention with a similar active control group increases
the validity of conclusions about the training program. Ideally,
researchers should compare training groups that differ only in
the demands of critical cognitive processes in order to disentangle
relevant components of a training intervention (cf. Schneiders,
Opitz, Krick, & Mecklinger, 2011). Likewise, if the training

2782

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

and transfer tasks share this critical process, the training is more
likely to produce learning transfer.
Previous research has demonstrated that the ability to control
interference is critical for WM and gF (Burgess, Gray, Conway, &
Braver, 2011; Gray, Chabris, & Braver, 2003; Kane, 2003). Interference control involves the detection and resolution of conflicts
that arise when irrelevant stimulus dimensions refer to an
incorrect stimulus or response pattern (Carter & van Veen, 2007;
Friedman & Miyake, 2004). In the n-back task, participants must
continuously track a stream of information and decide whether a
probe matches the stimulus presented n-steps back. In this task,
the process of interference control becomes crucial in lure trials
(Burgess et al., 2011; Gray et al., 2003; Kane, 2003), i.e. non-target
trials with a match in non-n positions. For instance, in a 2-back
task, the probe ‘P’ in the sequence ‘P–T–W–P’ is an n þ1 lure trial,
because the actual stimulus only corresponds with the stimulus
n þ1 trials before. Thus, n þ lures refer to items presented before
the nth item, whereas n  lures match with more recent stimuli.
These lures induce a familiarity signal which conflicts with explicit
recollection of the sequence (Oberauer, 2005; Szmalec,
Verbruggen, Vandierendonck, & Kemps, 2011). As a result, the
additional demand in interference control leads to increased
reaction times and false alarms rates. Specifically, lures that are
presented close to the critical lag (n þ1 or n  1) elicit the strongest
conflict (Kane, Conway, Miura, & Colflesh, 2007; McCabe &
Hartman, 2008; Szmalec et al., 2011). Thus, interference control
is a critical process in the n-back task, and training of this
component may lead to transfer in related cognitive domains.
Similar to isolating cognitive processes necessary for transfer,
mapping neural activity allows one not only to test whether
trained and untrained tasks overlap in specific brain regions but
also to estimate training induced activity changes (Dahlin et al.,
2008). A few studies have examined the brain regions that
mediate the relationship between working memory, interference
control and gF. In a functional magnetic resonance imaging (fMRI)
study using a visual 3-back task with n  1 and n þ 1/ þ2 lures,
Gray et al. (2003) determined that the blood oxygenation level
dependent (BOLD) signal in the left lateral prefrontal and bilateral
parietal regions explained 99.9% of the relationship between gF
and accuracy in lure trials. In trials with increased demand for
interference control, participants with high gF showed increased
signal in these regions. In another fMRI study, Burgess et al. (2011)
found that BOLD signal in the bilateral dorsolateral prefrontal
cortex (middle frontal gyrus) and the inferior parietal cortex
mediated the relationship between gF and WM. Thus, regions in
the lateral prefrontal and parietal cortices play a crucial role in
interference control (cf. Jonides & Nee, 2006; Kane, 2003; Nee,
Wager, & Jonides, 2007).
While fMRI allows for assessing the brain regions that contribute to transfer, event-related potentials (ERP) indicate the time
frames in which task-critical cognitive processes overlap. ERP
research on interference control has often focused on two components. The N2, a fronto-central negative component typically
elicited between 200 and 350 ms post-stimulus, is associated with
conflict monitoring or the resolution of conflict (Carter & van Veen,
2007; Kopp, Rist, & Mattler, 1996; Silton et al., 2010). The second
component, the parietal P3, which typically appears between 300
and 600 ms post-stimulus, is thought to reflect general processes
of attentional control, stimulus categorization, and the effort to
choose between competing stimuli or responses (Bledowski,
Prvulovic, Goebel, Zanella, & Linden, 2004; Neuhaus, Trempler
et al., 2010; Neuhaus, Urbanek et al., 2010; Rueda, Posner,
Rothbart, & Davis-Stober, 2004).
Based on the relationship between WM, gF, and interference
control, the present study was designed to investigate the
assumption that transfer effects result from partially shared neural

and cognitive mechanisms between the training and transfer
tasks. We tested this assumption on a behavioral and electrophysiological level with training and transfer tasks that tax
interference control heavily. We used the adaptive dual n-back
as a training task and manipulated the occurrence of lure trials
between training groups. To test for changes in interference
control, we used the Attention Network Test (ANT), a cued variant
of the flanker task (Fan, McCandliss, Sommer, Raz, & Posner, 2002).
Others have used this task (Fan, McCandliss, Fossella, Flombaum, &
Posner, 2005), both in healthy children and adults (Fan et al., 2002;
Rueda et al., 2004) and patients with various mental disorders
(Fernandez-Duque & Black, 2006; Neuhaus et al., 2010a, 2010b). In
the ANT, participants must quickly indicate the direction of a
central arrow, which is flanked by congruent or incongruent
arrows. Distinct cue-flanker combinations allow for assessing the
efficiency of three attentional networks. Conflict resolution is
manipulated by congruent and incongruent flanker arrows, and
the efficiency of alerting and orienting are derived from several
types of valid cues (see method section). Since we expected the
lure n-back training to increase the efficiency of conflict resolution,
which is thought to be closely related to interference resolution
(cf. Miyake et al., 2000; Szmalec et al., 2011), we focused on this
attention network for our theoretical and empirical analysis.
Our primary objective was to test whether training of shared
cognitive processes in the training and transfer tasks would lead to
improved performance in the ANT and corresponding electrophysiological changes. Researchers have found both increased and
decreased activations after task practice, and these changes may
arise from either implementing initial strategies more effectively,
learning new strategies during training or the reorganization of
cognitive functions (Dahlin et al., 2008; Jolles, Grol, Van Buchem,
Rombouts, & Crone, 2010). Unfortunately, the heterogeneity of
methods and tasks used in the field of cognitive training makes it
difficult to predict changes in neuronal activity precisely (cf.
Buschkuehl, Jaeggi, & Jonides, 2012). Nevertheless, based on the
aforementioned fMRI and ERP studies, we expected ERP effects in
the time range of the N2 and P3 components, with activation
focused in the fronto-parietal network. Specifically, we predicted
that we would detect changes in the time range where the training
and transfer tasks overlap in a topographic map of whole brain
activity. Additionally, we assessed WM capacity and gF to evaluate
transfer effects after training with the n-back task (Jaeggi et al.,
2008; Jaeggi et al., 2011; Studer-Luethi, Jaeggi, Buschkuehl, &
Perrig, 2012). Given the reported relationship between WM,
interference control, and gF, we predicted the largest transfer
gains after training with lure n-back.
2. Methods
2.1. Participants
Forty eight healthy, young adults, were recruited from an academic environment and randomly assigned to the lure training group, the non-lure training group
(i.e. the active control), or the passive control group. In each group, one participant
dropped out between pre- and post-testing and two dropped out during pretest
EEG recording, leaving a final sample of 43 participants (mean age: 25.2 years;
SD ¼4.1; range: 18–34). Both training groups consisted of 14 participants (6 female
and 1 left-handed each), and 15 people were assigned to the control group (8
female and 2 left-handed). Each individual submitted a written informed consent
before the experimental procedure began. After completion of the study, all
participants were paid 50 Swiss Francs for participation. The study was of the type
approved by the local ethics committee.
2.2. Materials
All training and transfer tasks were conducted with E-prime (Psychology
Software Tools, Inc., Pittsburgh, PA). We used a simplified version of the ANT in
pre- and post-testing (Fan et al., 2005), and modified the adaptive dual n-back task

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790
(Jaeggi et al., 2008) for the purpose of lure- and non-lure differentiation. Accuracies
and reaction times were recorded through a standard computer keyboard for the nback and a button box for the ANT. Offline ERP analyses were conducted using
BrainVision Analyzer (Brain Products, Munich, Germany) and Ragu (Koenig,
Kottlow, Stein, & Melie-García, 2011) for subsequent statistical ERP analyses.
2.3. Training and transfer tasks
2.3.1. Dual n-back (training)
Each training session consisted of 20 blocks ( E25 min), and each block
consisted of 20þ n trials, e.g. a 3-back block consisted of 23 trials. Each trial
presented a blue square at one of eight possible locations around a central fixation
cross. Meanwhile participants heard one of eight spoken letters (c, g, h, k, p, q,
t or w) via headphones. These visual and auditory stimuli were presented
simultaneously for 500 ms with an interstimulus interval (ISI) of 2500 ms. In both
modalities, participants had to press a key when the currently presented stimulus
matched the one presented previously. For example, in the 2-back, participants
needed to press ‘A’ with the left index finger when the letter spoken two stimuli
previously was the same as the current spoken letter. On visual match trials they
needed to press the ‘L’ key with the right index finger. Participants did not need to
press a key on non-target trials. After completion of one block, the program
presented feedback about the performance in both modalities and adjusted the nback difficulty level in an adaptive manner. If participants made more than
5 mistakes (i.e. misses and false alarms) in at least one modality, the n-level
dropped in the subsequent block. If they made fewer than 3 mistakes in both
modalities, the n-level increased by 1. Otherwise, they proceeded on the same nlevel in the next block. In sessions 1–3, daily training sessions began with a 1-back
block, and in sessions 4–14 with a 2-back block.
Irrespective of the current n-level, each block consisted of 6 targets per
modality. Out of 14 non-target trials, the lure n-back training contained an average
of 3 lure trials per block and modality, and we defined lures as probes that matched
the stimuli with a lag of n7 1 or n7 2. For instance, in a 3-back block, probes that
matched 1, 2, 4 or 5 trials back were lures. On the other hand, the non-lure n-back
training did not have any lures.
2.3.2. Pre- and post-assessments
2.3.2.1. Dual n-back. During pre- and posttest EEG recording, a total of 16 blocks
were administered. We used the lure n-back for the pre- and posttesting; however
unlike the training task, this task was non-adaptive and increased in difficulty from
2- to 5-back, with 4 blocks at each level of difficulty.
2.3.2.2. Attention Network Test. We used a simplified version of the ANT that contains 144 trials with equal numbers of three cue types and two flanker conditions
(Fan et al., 2005). Instructions were translated into German, and we adjusted the
display size of stimuli for EEG recordings to ensure comparable visual angles. The
program presented cues for 100 ms, which were displayed above, below (spatial
cues), or in place of the fixation cross (central cue), or else the cue was absent. After
a fixation period of 400 ms, an arrow pointing to the left or right was displayed for
1700 ms or until button press. The arrow appeared either above or below the fixation cross, and this arrow was flanked by two arrows on each side, pointing either
in the same (congruent) or the opposite direction (incongruent). Using their right
and left thumbs, participants aimed to quickly and accurately indicate the direction
of the central arrow while neglecting the four flankers. The interstimulus interval
(ISI) was variable such that ISI and RT summed up to 3500 ms (Fan et al., 2002).
2.3.2.3. Reading Span Task. To assess WM capacity, we used the Reading Span Task
with an automated procedure as described by Unsworth, Redick, Heitz, Broadway,
and Engle (2009). This task presented an alternating sequence of sentences and
capital letters, and participants first indicated whether the sentence made sense
and then were presented a letter to commit to memory. Once a sequence of 3 to 7
sentence-letter pairs has been displayed, participants had to reproduce the sequence of letters by clicking in the correct order on letters shown on screen. The
dependent variable was the sum of all correctly recalled letters (i.e. reading span
total). The total test duration was approximately 15–20 min.
2.3.2.4. Bochumer Matrizentest. The short version of the Bochumer Matrizentest –
advanced (BOMAT) was used to determine gF (Hossiep, Turck, & Hasella, 1999).
Similar to Raven's progressive matrices (Raven, 1990), this standardized test consists of 29 increasingly difficult visual analogy problems. We used parallel forms for
pre- and post-tests with the original time limit of 45 min. The total number of
correctly solved problems was the dependent variable.
2.4. Procedure
All participants performed the same tasks before and after an interval of 21
days to measure training induced effects. At the pretest session, we informed
participants about the experimental procedure and asked them to complete

2783

BOMAT and RST in a separate testing room. During EEG recording, they first
completed the ANT and then the n-back. At the end of the pretest, we informed
participants if they belonged to a training or the passive control group. Therefore,
participants in the passive control group knew that they belonged to the control
group of a training study, but participants in the two training groups were not
informed about the lure/non-lure manipulation. Both training groups trained with
the dual n-back for 14 self-administered sessions at home, distributed over 3 weeks,
with a weekly schedule of 5 consecutive training days and 2 days off. Individuals in
the control group were not contacted during the training interval. After this
interval, participants of all three groups were tested again with the tasks in the
same order (posttest).
2.5. ERP recording
The EEG measurement was conducted in an electrically shielded and airconditioned room at the Institute of Psychology, University of Bern. The viewing
distance was approximately 120 cm. We recorded from 64 Ag/AgCl electrodes
mounted in the 64 channel EasyCap electrode system (EASYCAP GmbH, Herrsching-Breitbrunn, Germany) in the extended 10–20 system. Two additional electrodes
placed below the left and the right eye recorded vertical eye movements, and
experimenters ensured impedances stayed below 20 kΩ. The recording reference
was at electrode position Fz, and two 32-channel BrainAmp MR plus amplifiers
(Brain Products, Munich, Germany) sampled the activity at 500 Hz and bandpassfiltered the boosted signal between .016 and 250 Hz.
In order to correct for eye movements, we used EEGLAB runica algorithms
(Delorme & Makeig, 2004) to perform an independent component analysis.
Components with a distinct temporal and spatial signature associated with eye
movements or blinks were then subtracted from the data. Then, in periods with
remaining artifacts, we used visual inspection to remove these time periods and
did not interpolate channels. The EEG data was then re-referenced to average
reference, downsampled to 250 Hz, and bandpass filtered at 0.3 Hz (24 db/oct)–
12.0 Hz (24 db/oct).
2.6. ERP analysis
The ERP analysis followed a similar procedure as described by Koenig and
Melie-García (2009). First, we computed grand mean ERPs of correct responses by
participant, session and condition (target locked, 0–800 ms, no baseline correction). Then, we applied topographic consistency tests for all time points to check if
grand averages across participants reflected a partially similar configuration of
generators (Koenig & Melie-García, 2010). Time points where the topographic
consistency test indicated that the probability of a consistent topography across
subjects was larger than .05 were excluded from further analyses. For the main
analysis, we used Topographic ANOVA (TANOVA) which is a non-parametric
randomization test based on global dissimilarities between electric fields
(Lehmann & Skrandies, 1980; Murray, Brunet, & Michel, 2008). As opposed to
more standard channel-wise statistics, a TANOVA results in a precise estimation of
the probability that the variance of global maps is related to the assumed structure
in the data (i.e. group, session, or condition). We first computed TANOVAs for each
consecutive time point, and in order to circumvent the problem of multiple testing,
we set a global duration cutoff (Koenig & Melie-García, 2009). The purpose of this
cutoff was to determine if the duration of a significant time-period exceeded
chance (Koenig et al., 2011).
Through our ERP analyses, we aimed to evaluate task overlap and training
induced effects. First, we conducted a TANOVA of the activity during the n-back and
the ANT to detect time frames that showed a conflict effect regardless of task and
testing session. Second, in a separate TANOVA, we assessed training induced effects
specific to the ANT. After observing periods above duration threshold, we
computed TANOVAs on the average electrical fields in order to assure that global
maps differed not just over single time points but actually over the whole period.
Post-hoc channel-wise t-tests (t-maps) and standardized low resolution brain
electromagnetic tomography (sLORETA) furthered our understanding of the topographic distribution and location of the neuronal generators of the map dissimilarities. We applied sLORETA based on 6239 equally spaced cortical gray matter
voxels. The forward solution was obtained using the boundary element method
applied to the MNI152 template (cf. Pascual-Marqui, 2002). A signal-to-noise ratio
(SNR) of 100 was assumed for the computation of the pseudoinverse matrix, but
pilot analyses with other SNRs yielded similar results.
2.7. Behavioral data analysis
An analysis of n-back training data was conducted to test for both improvements during training and differences between lure- and non-lure training. Moreover, we investigated if the lure training group improved their ability to correctly
reject lures. We also tested if lure training led to improvements in the ANT, RST,
and BOMAT.
Behavioral data were analyzed with analyses of variance (ANOVA), and for
pairwise comparisons, we conducted Tukey HSD (within subject) and Games-Howell

2784

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

(between subject) corrected tests (Howell, 2009). All reported p values resulted from
two-sided testing with an alpha of 0.05. Effect-sizes are reported with partial eta
squared (η2p ). For the sake of brevity, we focus only on significant results relevant to
our research questions.

3. Results
3.1. Behavioral results
3.1.1. n-back (training)
A total of four training sessions (1% of data) were missing and
thus linearly interpolated for subsequent training gain analysis.
One session was not registered due to technical problems and three
subjects completed 13/14 training sessions due to scheduling
issues. The n-back training groups reached an average n-level of
M¼2.16 (SD¼0.38) in the first session and an average of M¼ 3.78
(SD¼0.65) in the final session (Fig. 1). As expected, participants in
the lure training group reached a slightly lower n-level in their final
session. A repeated measures ANOVA on mean n-level with the
within-subjects factor session (average of 1st/2nd or 13th/14th) and
the between-subjects factor training group (lure n-back or non-lure
n-back) revealed a main effect of session, F(1, 26)¼216.56, po0.001,
η2p ¼0.89. There were no significant differences between the two
training groups, irrespective of session, although a trend was evident,
F(1, 26)¼ 3.76, p ¼ 0.06, η2p ¼ 0.13. The session by training group
interaction was not significant, F(1, 26)o1. Taken together, the
training groups improved in mean performance during training,
but they did not differ significantly in terms of training gain.
In order to test if participants in the lure training group
improved in rejecting lures, we performed a second analysis based
on false alarm rates (FA; i.e. average accuracies of both modalities
in non-target trials). First, a repeated measures ANOVA with the
within-subjects factors of trial type (lure or non-lure) and training
session (average of 1st/2nd or 13th/14th) revealed a significant
main effect of trial type indicating more lure trial FAs (F(1, 13) ¼
66.93, p o0.001, η2p ¼0.84; Fig. 2). The main effect of session
approached significance (F(1, 13) ¼4.08, p ¼0.06, η2p ¼0.24). Moreover, a significant type by session interaction (F(1, 13) ¼22.79,
p o0.001, η2p ¼ 0.64) indicated a significant reduction in lure trial
FAs (t(13) ¼3.21, p o0.01); however, in non-lure trials, we noted a
non-significant trend of FA increase from initial to final training
sessions (t(13) ¼  1.84, p¼ 0.09). Tukey HSD corrected pairwise
comparisons further indicated that in the initial training sessions,
n  2 lures had a significantly reduced rate of FAs compared to
n  1, n þ 1 (ps o0.05) and n þ2 lures (p o0.01), indicating a
processing advantage for recently displayed items. At the end of
the training, lure trials closer to the critical lag of n (i.e. n  1 and
n þ1) still yielded higher FA rates than n 2 lures, ps o0.01.
Interestingly, the improvement in nþ lures was significantly
higher than in n  lures, t(13) ¼3.16, p o0.01.
These results indicate that participants in the lure training
group improved their ability to reject lures, especially in the

Fig. 1. Average n-back training level by group (lure or non-lure training) as a
function of session. Error bars represent standard errors of the mean.

Fig. 2. n-back type of trial. Bars represent false alarm rates in non-target trials of
both modalities for non-lure and four different kinds of lure trials in sessions 1/2
and 13/14 of the lure n-back training. n  2 lures are closer to the current probe and
show a reduced false alarm rate, while n 71 lures close to the critical lag of n show
an increased rate irrespective of training session (see text). Error bars represent the
standard error of the mean.

Table 1
Descriptive data of pre-/post training assessment: Mean (SD).
Task

Group

Pre-test

Post-test

Lure
Non-lure
Control

0.33 (0.09)
0.31 (0.10)
0.33 (0.10)

0.72 (0.11)
0.68 (0.06)
0.40 (0.12)

Lure
Non-lure
Control
Lure
Non-lure
Control

481 (42)
469 (38)
477 (44)
562 (55)
569 (59)
572 (54)

440 (47)
452 (49)
468 (53)
513 (56)
522 (64)
548 (58)

Lure
Non-lure
Control

57.5 (10.5)
56.5 (9.4)
58.7 (8.4)

59.7 (10.8)
58.4 (9.1)
62.0 (5.3)

Lure
Non-lure
Control

15.3 (2.9)
15.7 (5.0)
16.5 (5.0)

18.5 (2.4)
18.9 (5.1)
19.9 (4.7)

n-back

ANT
Congruent

Incongruent

RST

BOMAT

Note. n¼ 43. n-back: accuracies (rates of hits – false alarms) in 2- to 5-back blocks.
ANT: Attention Network Test reaction times (ms) by congruency. RST: Automated
Reading Span Task, total sum of correctly recalled letters. BOMAT: Bochumer
Matrizen Test, total of correctly solved problems.

distant n þ lures. Lure trials led to higher FA rates compared to
non-lure trials, even at the end of training. Lures closer to the
critical lag (n  1 and n þ1) led to the highest FA rates.
3.1.2. Pre- and post-assessments
Descriptive data of pre- and post-assessment tests are displayed in Table 1. One-way ANOVAs with the between-subjects
factor group (lure n-back, non-lure n-back, or no training) did not
reveal any significant group differences at pretest, all Fs o 0.32,
ps 4 0.73.
3.1.2.1. Non-adaptive dual n-back. We calculated pr values (hit rate
– false alarms rate) to compare performance of the non-adaptive
dual n-back in pre- and post-test. An ANOVA with the betweensubjects factor group (lure n-back, non-lure n-back, or no training)
and the within-subjects factor session (pre- or post-test) revealed
a main effect of session, F(1, 40) ¼509.56, p o0.001, η2p ¼0.93, and a
significant session by group interaction, F(2, 40) ¼81.98, p o0.001,
η2p ¼0.80. Importantly, when the analysis was restricted to the two
training groups, the critical session by group interaction was not
significant, F(1, 26) o1, p ¼ 0.59, η2p ¼0.01, and the main effect
group did not reach level of significance, F(1, 26) ¼1.02, p ¼ 0.32,

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

η2p ¼0.04. Therefore, although the non-lure training group tended
to reach higher levels during training, performance in the nonadaptive lure n-back did not differ between the two training groups.
3.1.2.2. Attention Network Test. Overall, participants performed
with very high accuracy in the ANT and higher in the congruent
(MACC ¼0.99, SDACC ¼0.02) compared to the incongruent condition
(MACC ¼0.94, SDACC ¼0.05). There were no reliable changes associated with group. Therefore, the focus of the ANT analysis was on
reaction times (RT). We analyzed RTs within a range of
200–1000 ms (2.3% of data excluded), and we used correct
responses for further analysis. RTs were shorter in the congruent
condition (MRT ¼465 ms, SDRT ¼44) compared to the incongruent
condition (MRT ¼548 ms, SDRT ¼ 55; Table 1).
A repeated measures ANOVA on reaction times with the
between-subjects factor group (lure n-back, non-lure n-back, or
no training) and the within-subjects factors session (pre- or posttest), congruency (congruent or incongruent) and cue (spatial,
central, or none) revealed a main effect of session (F(1, 40)¼53.51,
po0.001, η2p ¼0.57). This indicated that RTs were significantly
shorter in the post-test, and a session by congruency interaction
(F(1, 40)¼29.35, po0.001, η2p ¼0.42) showed that these RT reductions were more pronounced in the incongruent condition.1 Importantly, there was also a significant group by session by congruency
interaction, F(2, 40)¼3.97, po0.05, η2p ¼ 0.17. We decomposed this
three-way interaction into two ANOVAs, separated by congruency.
In the congruent condition only, the session by group interaction
was significant (congruent: F(2, 40)¼ 6.00, po0.01, η2p ¼0.23;
incongruent: F(2, 40)¼2.66, p¼ 0.08, η2p ¼0.12). In this condition,
the lure training group improved significantly more than both the
non-lure (t(26)¼2.68, po0.05) and the control group (t(27)¼ 3.57,
po0.01), whereas the non-lure and the control group did not differ
(t(27)¼0.76, p¼0.73; Games-Howell corrected). Finally, regardless
of congruency and cue, a significant session by group interaction
(F(2, 40)¼3.96, po0.05, η2p ¼0.16) indicated group-specific RT
changes. However, this interaction was mainly driven by the
significant session by group interaction found in the congruent
condition. Specifically, as indicated by descriptive data (cf. Table 1),
there were no reliable differences between the lure and the nonlure training group in the incongruent condition.
In summary, RTs decreased significantly from pre- to post-test,
and this decrease was more pronounced in the incongruent
condition regardless of group. While there was a tendency towards
differential improvements in the incongruent condition, we only
found reliable group differences in the congruent condition. In this
condition, the lure training group improved significantly more
than the other two groups.
3.1.2.3. Reading Span Task and Bochumer Matrizentest. In the RST,
the total number of correctly recalled letters was M ¼57.6
(SD ¼ 9.3) in the pretest and M ¼60.1 (SD ¼ 8.6) in the post-test.
A repeated measures ANOVA with the between-subjects factor
group (lure n-back, non-lure n-back, or no training) and the
within-subject factor session (pre- or post-test) revealed no
significant main effect of group, no interaction (both Fs o1), and
only a statistical trend for session, F(1, 40) ¼3.35, p ¼ 0.07,
η2p ¼0.08. In the BOMAT, the number of correctly solved analogy
problems increased from M ¼15.9 (SD ¼4.3) in the pretest to
M ¼19.1 (SD ¼4.2) in the post-test, which was confirmed by a
significant main effect of session, F(1, 40) ¼38.24, p o0.001,
1
The main effects for congruency (F(1, 40) ¼ 440.45, p o 0.001, η2p ¼ 0.92), cue
(F (2, 80) ¼295.23, p o 0.001, η2p ¼ 0.88) and a significant interaction between these
factors (F(2, 80)¼ 17.11, p o 0.001, η2p ¼ 0.30) are task inherent and were not further
analyzed. Besides the results reported in the text, no other main effects or
interactions were significant.

2785

η2p ¼0.49. However, the main effect of group and the relevant
session by group interaction did not reach significance, Fs o1. Also,
combining the two training groups and comparing them to the
passive control group did not reveal a main effect of group or a
group by session interaction for RST and BOMAT (all Fs o1). Thus,
the RST and BOMAT scores were higher in the posttest, but neither
the lure training nor the non-lure training group showed a higher
pre–post gain compared to the control group.
3.2. ERP results
3.2.1. Overlap analysis
The post-stimulus ERPs are displayed in Fig. 3a. In the n-back
task, lure and non-lure trials displayed a strong fronto-central
positivity from 150 to 300 ms and a parietal P3 complex from 200
to 600 ms. We also detected a frontal positive component between
150 and 300 ms in the ANT, which was also followed by a
pronounced parietal P3 component. In order to investigate the
overlap of components in the n-back and ANT, we conducted a
repeated measures TANOVA with the within-subject factors task
(n-back or ANT) and conflict (lure/incongruent or non-lure/congruent). The main effect of task was significant within the entire
time range tested (0–800 ms). Thus, the two tasks clearly differed
in topography. Interestingly, we observed a significant main effect
of conflict within 332–704 ms and a significant task by conflict
interaction within 344–444 and 564–680 ms (Fig. 3b). Although
the conflict effect was more pronounced in the ANT, both tasks
showed a similar conflict effect within 332–704 ms, with a
positive centroid moving from the anterior to posterior electrodes
(see the difference maps in Fig. 3c). This overlap was most clear in
the range of 444–564 ms. Interestingly, in this time period with a
significant conflict effect, there was no interaction of task and
conflict.
These results indicate that although the two tasks do differ,
there is a substantial overlap in topography of the conflict effect in
the P3 range from 444 to 564 ms, and the absence of an interaction indicates that there is no evidence for a difference of this
conflict effect between the two tasks. This suggests that from 444
to 564 ms similar activity may underlie both types of interference
control, and lure-training has induced changes in this time
window that have an effect on conflict processing in the ANT task.
Hence, transfer is most likely to be promoted by the processes
underlying the P3 component and common for both tasks within
this time period.
3.2.2. Transfer analysis
In order to analyze training effects on ERPs in the ANT, we
computed a repeated measures TANOVA with the betweensubjects factor group (lure n-back, non-lure n-back, or no training)
and the within-subjects factors session (pre- or posttest) and
congruency (congruent or incongruent). We found several significant effects: an effect of congruency from 260 to 800 ms, a main
effect of session within 0–196 ms, 212–488 ms, and 564–764 ms,
and a group by session interaction within 300–564 ms. No other
main effect or interaction reached significance. Since only the
group by session interaction was relevant to our research questions, this became the focus of our subsequent analyses.2
We computed a group by session TANOVA on averaged activity
in the interval within 300–564 ms, and again, the interaction
reached significance, po 0.001. Subsequent pairwise comparisons
2
A separate TANOVA with the factors group, session, and cue (spatial, central,
or none) revealed a significant main effect of cue within the full range (0–800 ms)
but no significant interactions (session by group or session by group by cue)
relevant to the current study.

2786

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

Fig. 4. ANT training effect t-Maps/LORETA. Attention Network Test t-maps reflecting session effect (post- minus pretest). Session differences are displayed for the
lure training and control groups (non-lure training and passive control group
combined) within 300–564 ms post-stimulus, as well as the difference between the
two, reflecting the interaction of session (post- minus pretest) and group (lure
training minus control groups). This difference reflecting the session by group
interaction is also reflected in the LORETA, which indicates training induced activity
increases for the lure training group. Activation display is thresholded at ǀtǀZ 2.02
(p o 0.05, uncorrected).

Fig. 3. (A) Post-stimulus ERPs. Grand-averaged ERPs are displayed in selected
channels for lure and non-lure trials in the n-back task (left) and incongruent and
congruent trials in the ANT (right). Gray bars indicate TANOVA time points with a
significant main effect of conflict, indicating partially overlapping topography
between the two tasks (see text). (B) Results of the TANOVA task by conflict. pvalues are displayed for main effect of conflict and the task by conflict interaction.
The gray line represents a threshold of p ¼0.05. Whereas the main effect of conflict
is significant for all time points within this period, the significant interaction within
344–444 ms and 564–680 ms suggests partly differential conflict effects in the two
tasks. (C) Maps of electrophysiological activity are shown for incongruent/congruent trials in ANT and lure/non-lure trials in n-back. Difference maps reflecting conflict effects in both tasks suggest considerable overlap in the time frame
of 444–564 ms.

between groups yielded significant differences between the lure
training and the other groups (ps o .001), but we did not discover
differences between the non-lure training and control group
(p ¼0.15). In order to clarify the localization of the pre/postdifferences between groups, we calculated post- minus pretest tmaps for the lure training group and compared them to the
combined non-lure training and control group (Fig. 4). The tmap reflecting the group by session interaction revealed increased
fronto-central positivity (tmax ¼ 2.99 at FC4) and posterior negativity (tmin ¼  4.07 at Iz) for the lure training group, compared to
pre-/post-test differences for the control groups. Finally, we
computed a LORETA on the posttest data using the pretest data
as a baseline. This allowed us to reveal tomographic activity
changes after lure training compared to pre-/post-test differences
in the non-lure and control group (Fig. 4). The LORETA revealed
three clusters with increased activation for the lure training group.
74% of all the activated voxels above threshold (tZ 2.02) were
located within a large cluster that compromised the postcentral

gyrus (BA 5/7/40), precuneus (BA 7), and the superior (BA 7) and
inferior parietal lobule (BA 40), which was also the source of
maximal activation (BA 40; t¼ 2.76). Additionally, two small
clusters centered in the right hemisphere in superior temporal
gyrus (BA 41; t¼2.28) and middle occipital gyrus (BA 19; t¼2.35)
reached significance. There was no indication of significant
decreases in activation level.
Interestingly, this session by group interaction (300–564 ms)
coincided with the main effect of conflict that we found in the
electrophysiological overlap analysis (332–704 ms), which
includes the time frame of the conflict effect that did not coincide
with the task by conflict interaction (444–564 ms). A separate
TANOVA on ANT ERPs indicated that the session by group interaction was also significant within this latter period (444–564 ms).
Within these time windows, the lure training group differed
significantly from the non-lure training (po 0.01) and control
group (p o0.05), yet again, the non-lure training and the control
group did not differ from each other (p ¼0.39).
In summary, the first analysis showed overlapping topography
between the n-back task and the ANT in a time range 332–704 ms,
and this overlap was more pronounced within 444–564 ms poststimulus. This difference of incongruent/lure trials and congruent/
non-lure trials showed a positive fronto-central component followed by a parietal positivity. In a second analysis, the session by
group interaction indicated that change in ERPs in the lure training
group differed significantly from the one in the non-lure training
and control group. Using LORETA, we analyzed these differences
further and noted an increased activity in left parietal scalp sites
after the lure training.

4. Discussion
The present study investigated behavioral and electrophysiological effects following 14 sessions of adaptive dual n-back
training. To estimate improvements in interference control, we
compared performance of two groups training with lure and nonlure variants of the adaptive dual n-back with a passive control
group. Lure trials were defined as non-targets with a match within

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

the range of 72 items around the critical lag of n. In order to
assess the potential benefits of improved interference control, we
administered three tasks measuring control of attention, working
memory, and fluid intelligence. We observed task-specific
improvements after n-back training, yet behavioral measures of
transfer revealed unexpected results. In the ANT, we expected
conflict resolution to improve most after lure n-back training.
However, lure training led to faster responses only in the congruent condition, while participants improved more in the incongruent condition regardless of group. Interestingly, we did not note
any differential effects in the Reading Span Task and BOMAT, i.e.
neither type of n-back training led to transfer in these tasks
measuring WM and gF. By contrast, ERP analyses revealed unique
lure training induced topographical differences in the P3 range,
while we observed no differences between the non-lure training
and the control group. Using LORETA, we localized these activations to the left parietal lobule, which displayed higher activity
after training with lure n-back. In a separate analysis, we determined that conflict induced by the lures in n-back task and the
incongruent flankers in the ANT resulted in similar ERPs, although
we noted obvious differences between these tasks as well. This
suggests that there may exist some similarities in the neural
response to lures that produce proactive interference and the
neural response to the presence of conflict in the flanker task. This
similarity may reflect neural processes in the left parietal cortex
common to both conditions, which would provide a basis for
transfer between these tasks.
A prerequisite for transfer is improvement on the training task;
further, transfer of interference control would specifically require
improvement in the rejection of lures in the n-back task. Therefore, we expected increased average n-levels for both training
groups and, more specifically, an improved ability to reject lures
for the lure n-back training group. Both training groups improved
during training, but the average n-level was slightly higher during
non-lure training. One could argue that the higher difficulty in the
lure training contributed to the observed electrophysiological and
behavioral effects. However, as the n-level adjusted to individual
performance level in both n-back tasks, we assume that overall
difficulty did not differ. This assumption is supported by the
results of a post-session questionnaire indicating that participants
in both groups perceived the tasks as equally challenging.3 Therefore, it is rather unlikely that differences in task difficulty can
explain such group differences. More importantly, during the
training, the lure n-back group decreased their rate of false alarms
in lure trials, whereas false alarms of non-lure trials were
constantly low. Interestingly, participants' rate of false alarms
did decrease more in distant n þ than in close n  lures. This
might indicate that they learned to distinguish between items in
n  positions that are still needed for further processing and n þ
items that are no longer relevant (Szmalec et al., 2011). This could
be interpreted as an increase in efficiency of updating working
memory, or may be due to an overall increase in speed of
processing in the lure training group. Generally, we observed the
highest false alarm rates in lure trials close to the critical lag (n 1/
n þ1). This indicates the participants' sensitivity to interference in
updating tasks since weak bindings of items with corresponding
serial positions may hinder counteracting familiarity-based decisions (Kane et al., 2007; Schmiedek, Li, & Lindenberger, 2009;
Szmalec et al., 2011).

3
At the beginning of the posttest, subjects in the training groups had to
indicate on a scale from 1 to 9 how much effort and concentration the training
required. T-tests indicated that the mean rating for both items did not differ
between groups (effort: lure group: M ¼6.14, SD ¼1.83, non-lure group: M¼ 6.71,
SD ¼ 1.73; t(26) ¼0.85, p ¼ 0.40; concentration: lure group: M ¼8.36, SD ¼ 1.01, nonlure group: M ¼8.57, SD ¼ 0.65; t(26) ¼0.67, p ¼ 0.51).

2787

The major improvement in rejecting n þ lures indicates that
participants in the lure training group in fact learned to distinguish between relevant items and no longer relevant but still
interfering items. Interestingly, it has been suggested that in tasks
where interference trials are frequent and therefore can be
anticipated, interference control is active prior to conflict onset
(i.e. proactive control; Braver, Gray, & Burgess, 2007; Burgess &
Braver, 2010; Burgess et al., 2011). Also, Mevorach, Humphreys,
and Shalev (2009) showed that the left posterior parietal cortex is
involved in presetting a system to ignore salient distractors or to
preselect a window of attention. Because of the high interference
expectancy in the ANT (50%), a shift from reactive to proactive
control would explain the observed improved performance irrespective of interference onset. Moreover, it is interesting to see
that the region around the left intraparietal sulcus, which was
more active after lure training, has been associated with proactive
or preparatory attentional processes.
While improvements on the training task are necessary for
transfer to occur, they are not sufficient to prove an enhancement
in ability. Improvements of training performance may also reflect
newly developed strategies specific to the training task, learned
and automated stimulus-response configurations or other unintended task-specific changes (cf. Chooi & Thompson, 2012;
Klingberg, 2010; Shipstead et al., 2010). Consequently, although
the lure n-back and the ANT do not seem to share many task
features, we expected enhanced conflict resolution in the ANT on
the assumption that processes engaged in both tasks partially
overlap. However, some reports indicate that resolution of proactive interference in n-back and conflict resolution in flanker tasks
engage different inhibition processes (cf. Bissett, Nee, & Jonides,
2009; Nee et al., 2007). By contrast, Szmalec et al. (2011) suggest
that interference in n-back involves similar mechanisms as in
several response-congruency tasks. Our overlap analysis of ERPs in
the n-back task and ANT showed similar conflict effects in the P3
range (332–704 ms). One must note that this data-driven ERP
analysis does not prove or quantify the degree of electrophysiological overlap, but is instead suggestive of a similarity in neuronal
activity and cognitive processes required in both tasks. More
importantly, the analysis is primarily informative about the
relevant time range where tasks overlap and indicates which
processes may promote transfer.
The analysis of training induced electrophysiological changes in
the ANT yielded reliable group differences in the P3 time range
(300–564 ms post-stimulus). Compared to active and passive
control group, the lure training led to increased activity in the
inferior and superior parietal lobule in the left hemisphere. By
contrast, 14 sessions of n-back training without lures did not
reveal any electrophysiological effects compared to the passive
control group. Such activation differences can be interpreted based
on previous studies, showing that the regions around the intraparietal sulcus support the voluntary control of attention, including shifts of attention (Hopfinger, Buonocore, & Mangun, 2000;
Walsh, Buonocore, Carter, & Mangun, 2011).
Taken together, the analyses on ERPs suggest that overlap and
transfer may entail more generic attentional effects, and not
interference control specifically. First, the overlap analysis did
not include the earlier N2 component, which has been associated
with conflict processing and possibly has a neural generator in the
anterior cingulate cortex (ACC; Carter & van Veen, 2007; Kopp
et al., 1996; Silton et al., 2010). Therefore, although n-back lures
and incongruent flankers have been associated with activation in
the ACC together with other prefrontal regions (Burgess et al.,
2011; Fan et al., 2005; Gray et al., 2003; Nee et al., 2007), the tasks
may differ in respect of the time course of conflict monitoring or
the resolution of a conflict. Second, given the large time period
reflecting overlapping conflict effects that exceeds response times

2788

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

in the ANT, it is most likely that these similar activations induced
by the conflict manipulation reflect more general aspects of
cognitive or attentional control and not the detection of a conflict
or the action of inhibition. Third, the LORETA we conducted
revealed activation differences in the lure group which were
restricted to posterior, mainly left parietal sites, and did not
include any frontal regions. Therefore, the electrophysiological
data do not suggest specific effects of lure n-back training on
interference control, but the training may have improved attentional control.
While lure training induced electrophysiological changes irrespective of congruency, it only had a behavioral effect on the
congruent condition of the ANT. In general, one could argue that
the reduced reaction times simply occurred because of an increased
speed of processing, similar to that observed after training with
action video games (Dye, Green, & Bavelier, 2009). It is assumed that
in n-back tasks without lures, performance is mainly based on
familiarity matching, requiring rather little cognitive control
(Szmalec et al., 2011). By contrast, in the lure n-back task, cognitive
control is constantly needed to maintain task goals, prioritize
recollection over familiarity and constrain attention only to the
relevant items, i.e. the last n presented items. Therefore, these
subjects may have learned that they need to make faster decisions
after the stimulus presentation. E.g. if they need 1 s to decide
whether the current stimulus is a target or a non-target, there are
2 s left till the next stimulus presentation to rehearse n stimuli in
each of the two modalities (cf. Barrouillet, Bernardin, & Camos,
2004). This would result in faster reaction times in the ANT
irrespective of congruency, but would not lead to improved performance in RST and BOMAT, which are less sensitive for stimulus
processing.4 However, it remains an open question whether the
shortened RTs for the lure training group in the n-back task and the
ANT reflect an improvement of a cognitive ability or a simple strategy
shift. In general, it has been argued that strategies tend to be limited
to a particular task or type of information (cf. Morrison & Chein,
2011). On the one hand, it is very likely that subjects in the lure
training group developed a different strategy during 14 days of nback training. On the other hand, the application of a new n-back
strategy does not seem to explain shortened RTs in the ANT. At least a
simple strategy to make faster decisions would likely lead to
decreased reaction times and to decreased accuracies (i.e. speedaccuracy trade-off). However, accuracies were generally high and did
not deteriorate more in the lure training group. Therefore, it seems
rather unlikely that subjects training with lures specifically transferred a learned strategy from the n-back task to the ANT, but our
study provides no definite answer to that question.
Given the reported relationship between WM, interference
control, and gF and the earlier reports of far transfer observed
after training with the adaptive n-back task (Jaeggi et al., 2008,
2011; Studer-Luethi et al., 2012), we expected training induced
improvements in measures of WM and gF. However, we found no
reliable group differences in either task. One interpretation of this
finding is that the effects were too small to promote reliable far
transfer effects. Furthermore, WM and gF may be stable individual
characteristics that are not easily modified with a few weeks of
cognitive training. A recent report by Redick et al. (2012) seems to
confirm this conclusion. In their study, 20 sessions of adaptive dual
n-back training did not lead to any substantial performance
improvements in several tasks measuring fluid and crystallized

4
This interpretation is backed up by a post-hoc analysis of average reaction
times during lure and non-lure training. Average reaction times of hits in the initial
two sessions did not differ between groups (lure group: M ¼ 1187 ms, SD ¼179 ms;
non-lure group: M ¼1228 ms, SD ¼ 231ms; t(26)¼ 0.53, p ¼ 0.60), but differed
significantly in the last two sessions (lure group: M¼ 836 ms, SD¼ 169 ms; nonlure group: M ¼1000 ms, SD ¼ 230 ms; t(26) ¼2.16, p o 0.05).

intelligence, WM, multitasking, and perceptual speed. Altogether,
we conclude that overlapping cognitive processes may be a
requirement for transfer to occur, but the overlap itself is obviously
not sufficient. Nevertheless, it is still questionable if working
memory and fluid intelligence are trainable at all.
While our study was designed to increase the validity of conclusions about the training program, some important limitations should
be noted. We used an n-back paradigm which places heavy demands
on inference control and expected to find improved interference
control and transfer to related domains. The lure manipulation
elicited the expected results with increased false alarm rates around
the critical lag of n, and the lure training group was able to improve
the lure rejection during training. As mentioned before, the electrophysiological and behavioral data do not suggest that inference
control improved in the lure training group, but can be interpreted
as either an improvement in attentional control or processing speed.
Importantly, we included only one task that measured the cognitive
ability we attempted to train (i.e. interference control). In order to
minimize task specific variance and reduce the influence of shared
task features that are common to training and transfer tasks, transfer
should be assessed on the level of latent variables which were
derived from several tasks measuring the same underlying construct
(Morrison & Chein, 2011; Shipstead, Redick, et al., 2012). Moreover,
the design of the current study did not permit to compare lure/nonlure performance between the two training groups. In order to keep
the testing sessions reasonably short, the n-back task did not include
enough trials to assess reliable accuracy data on the level of lures and
non-lures, and this is a clear limitation of the current study. While in
our task, key presses were only necessary for n-back matches, future
studies should choose a paradigm where subjects need to press a key
in mismatch trials as well. This would allow for assessing precisely
the effect of the lures on the reaction times in current and
subsequent trials (Szmalec et al., 2011). Furthermore, we included a
non-adaptive n-back in order to run the same experimental procedure for all subjects and testing sessions. However, without training,
the higher n-levels are very difficult and may have led to decreased
motivation and lower reliability. Future studies should therefore
include reliable neural and behavioral indices of training and transfer
that could also be compared between the experimental groups.
With the current study, we addressed a critique by Morrison and
Chein (2011), who note that most training studies did not control
variables such as training motivation or task difficulty, which may
confound with different training regimens. For example, participants
training with adaptive as opposed to non-adaptive tasks may be
more convinced that this task will result in cognitive improvement
(cf. Shipstead, Hicks, & Engle, 2012; Shipstead, Redick, et al., 2012).
Therefore, placebo effects cannot be controlled completely with nonadaptive training tasks. Furthermore, we address critiques from Boot,
Blakely, and Simons (2011), as the participants in our study had no
way of knowing whether they belonged to the experimental or active
control group. Since we administered two adaptive dual n-back
variants, which differed only in demand of a single executive core
process, we believe that the noted differences between the two
training groups can clearly be traced back to this manipulation. Also,
though we found electrophysiological and behavioral differences in
the ANT between the lure n-back training and the control group,
there was no indication of any reliable difference between the nonlure training and the control group. We therefore believe that future
studies should pursue this methodological approach in order to
obtain valid results about the trainability of WM and related abilities.

Acknowledgments
We thank Karin Bruegger, Beatrice Rumpel and Tina Schneider
for help with data collection. We also would like to thank Marco

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

Hollenstein and two anonymous reviewers for their helpful
comments and suggestions.

References
Barrouillet, P., Bernardin, S., & Camos, V. (2004). Time constraints and resource
sharing in adults' working memory spans. Journal of Experimental Psychology:
General, 133, 83–100.
Bissett, P. G., Nee, D. E., & Jonides, J. (2009). Dissociating interference-control
processes between memory and response. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 35, 1306–1316, http://dx.doi.org/10.1037/
a0016537.
Bledowski, C., Prvulovic, D., Goebel, R., Zanella, F. E., & Linden, D. E. J. (2004).
Attentional systems in target and distractor processing: A combined ERP and
fMRI study. Neuroimage, 22, 530–540, http://dx.doi.org/10.1016/j.neuroimage.
2003.12.034.
Boot, W. R., Blakely, D. P., & Simons, D. J. (2011). Do action video games improve
perception and cognition? Frontiers in Psychology, 2, 1–6, http://dx.doi.org/
10.3389/fpsyg.2011.00226.
Braver, T. S., Gray, J. R., & Burgess, G. C. (2007). Explaining the many varieties of
working memory variation: Dual mechanisms of cognitive control. In: A. R.
A. Conway, C. Jarrold, M. J. Kane, A. Miyake, & J. Towse (Eds.), Variation in
working memory. Oxford: Oxford University Press.
Burgess, G. C., & Braver, T. S. (2010). Neural mechanisms of interference control in
working memory: Effects of interference expectancy and fluid intelligence. PLos
One, 5, e12861, http://dx.doi.org/10.1371/journal.pone.0012861.
Burgess, G. C., Gray, J. R., Conway, A. R., & Braver, T. S. (2011). Neural mechanisms
of interference control underlie the relationship between fluid intelligence
and working memory span. Journal of Experimental Psychology: General, 140,
674–692, http://dx.doi.org/10.1037/A0024695.
Buschkuehl, M., Jaeggi, S. M., & Jonides, J. (2012). Neuronal effects following
working memory training. Developmental Cognitive Neuroscience, 2(Suppl. 1),
S167–S179, http://dx.doi.org/10.1016/j.dcn.2011.10.001.
Carter, C. S., & van Veen, V. (2007). Anterior cingulate cortex and conflict detection:
An update of theory and data. Cognitive Affective & Behavioral Neuroscience, 7,
367–379.
Chooi, W.-T., & Thompson, L. A. (2012). Working memory training does not improve
intelligence in healthy young adults. Intelligence, 40, 531–542, http://dx.doi.
org/10.1016/j.intell.2012.07.004.
Dahlin, E., Stigsdotter Neely, A., Larsson, A., Bäckman, L., & Nyberg, L. (2008).
Transfer of learning after updating training mediated by the striatum. Science,
320, 1510–1512, http://dx.doi.org/10.1126/science.1155466.
Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of
single-trial EEG dynamics including independent component analysis. Journal
of Neuroscience Methods, 134, 9–21, http://dx.doi.org/10.1016/j.jneumeth.
2003.10.009.
Detterman, D. (1993). The case for the prosecution: Transfer as an epiphenomenon.
In: D. Detterman, & R. Sternberg (Eds.), Transfer on trial: Intelligence, cognition,
and instruction (pp. 1–24). Norwood, NJ: Ablex.
Dye, M. W., Green, C. S., & Bavelier, D. (2009). Increasing speed of processing with
action video games. Current Directions in Psychological Science, 18, 321–326,
http://dx.doi.org/10.1111/j.1467-8721.2009.01660.x.
Fan, J., McCandliss, B. D., Fossella, J., Flombaum, J. I., & Posner, M. I. (2005). The
activation of attentional networks. Neuroimage, 26, 471–479, http://dx.doi.org/
10.1016/j.neuroimage.2005.02.004.
Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the
efficiency and independence of attentional networks. Journal of Cognitive
Neuroscience, 14, 340–347.
Fernandez-Duque, D., & Black, S. E. (2006). Attentional networks in normal aging
and Alzheimer's disease. Neuropsychology, 20, 133–143, http://dx.doi.org/
10.1037/0894-4105.20.2.133.
Friedman, N. P., & Miyake, A. (2004). The relations among inhibition and interference control functions: a latent-variable analysis. Journal of Experimental
Psychology: General, 133, 101–135.
Gray, J. R., Chabris, C. F., & Braver, T. S. (2003). Neural mechanisms of general fluid
intelligence. Nature Neuroscience, 6, 316–322.
Hopfinger, J. B., Buonocore, M. H., & Mangun, G. R. (2000). The neural mechanisms
of top-down attentional control. Nature Neuroscience, 3, 284–291.
Hossiep, R., Turck, D., & Hasella, M. (1999). Bochumer Matrizentest. BOMAT –
advanced – short version. Hogrefe: Göttingen.
Howell, D. C. (2009). Statistical methods for psychology (7th edition). Belmont, CA:
Wadsworth.
Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Improving fluid
intelligence with training on working memory. Proceedings of the National
Academy of Sciences, USA, 105, 6829–6833.
Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Shah, P. (2011). Short- and long-term
benefits of cognitive training. Proceedings of the National Academy of Sciences,
USA, 108, 10081–10086, http://dx.doi.org/10.1073/pnas.1103228108.
Jolles, D. D., Grol, M. J., Van Buchem, M. A., Rombouts, S. A. R. B., & Crone, E. A.
(2010). Practice effects in the brain: Changes in cerebral activation after
working memory practice depend on task demands. Neuroimage, 52, 658–668,
http://dx.doi.org/10.1016/j.neuroimage.2010.04.028.
Jonides, J. (2004). How does practice makes perfect? Nature Neuroscience, 7, 10–11.

2789

Jonides, J., & Nee, D. E. (2006). Brain mechanisms of proactive interference in
working memory. Neuroscience, 139, 181–193, http://dx.doi.org/10.1016/j.
neuroscience.2005.06.042.
Kane, M. J. (2003). The intelligent brain in conflict. Trends in Cognitive Sciences, 7,
375–377, http://dx.doi.org/10.1016/S1364-6613(03)00193-1.
Kane, M. J., Conway, A. R. A., Miura, T. K., & Colflesh, G. J. H. (2007). Working
memory, attention control, and the N-back task: A question of construct
validity. Journal of Experimental Psychology: Learning, Memory and Cognition,
33, 615–622, http://dx.doi.org/10.1037/0278-7393.33.3.615.
Klingberg, T. (2010). Training and plasticity of working memory. Trends in Cognitive
Sciences, 14, 317–324, http://dx.doi.org/10.1016/j.tics.2010.05.002.
Klingberg, T., Fernell, E., Olesen, P. J., Johnson, M., Gustafsson, P., Dahlstrom, K., et al.
(2005). Computerized training of working memory in children with ADHD-A
randomized, controlled trial. Journal of the American Academy of Child &
Adolescent Psychiatry, 44, 177–186.
Koenig, T., Kottlow, M., Stein, M., & Melie-García, L. (2011). Ragu: A free tool for the
analysis of EEG and MEG event-related scalp field data using global randomization statistics. Computational Intelligence and Neuroscience, 2011, 938925, http:
//dx.doi.org/10.1155/2011/938925.
Koenig, T., & Melie-García, L. (2009). Statistical analysis of multichannel scalp field data.
In: C. M. Michel, T. Koenig, D. Brandeis, L. R.R. Gianotti, & J. Wackermann (Eds.),
Electrical Neuroimaging (pp. 169–190). Cambridge: Cambridge University Press.
Koenig, T., & Melie-García, L. (2010). A method to determine the presence of
averaged event-related fields using randomization tests. Brain Topography, 23,
233–242, http://dx.doi.org/10.1007/s10548-010-0142-1.
Kopp, B., Rist, F., & Mattler, U. (1996). N200 in the flanker task as a neurobehavioral
tool for investigating executive control. Psychophysiology, 33, 282–294.
Lehmann, D., & Skrandies, W. (1980). Reference-free identification of components
of checkerboard-evoked multichannel potential fields. Electroencephalography
and Clinical Neurophysiology, 48, 609–621.
McCabe, J., & Hartman, M. (2008). Working memory for item and temporal
information in younger and older adults. Aging, Neuropsychology and Cognition,
15, 574–600, http://dx.doi.org/10.1080/13825580801956217.
Mevorach, C., Humphreys, G. W., & Shalev, L. (2009). Reflexive and preparatory
selection and suppression of salient information in the right and left posterior
parietal cortex. Journal of Cognitive Neuroscience, 21, 1204–1214, http://dx.doi.
org/10.1162/jocn.2009.21088.
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D.
(2000). The unity and diversity of executive functions and their contributions
to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology,
41, 49–100.
Morrison, A. B., & Chein, J. M. (2011). Does working memory training work? The
promise and challenges of enhancing cognition by training working memory.
Psychonomic Bulletin & Review, 18, 46–60, http://dx.doi.org/10.3758/s13423010-0034-0.
Murray, M. M., Brunet, D., & Michel, C. M. (2008). Topographic ERP analyses: A stepby-step tutorial review. Brain Topography, 20, 249–264, http://dx.doi.org/
10.1007/s10548-008-0054-5.
Nee, D. E., Wager, T. D., & Jonides, J. (2007). Interference resolution: Insights from a
meta-analysis of neuroimaging tasks. Cognitive Affective & Behavioral Neuroscience, 7, 1–17.
Neuhaus, A. H., Trempler, N. R., Hahn, E., Luborzewski, A., Karl, C., Hahn, C., et al.
(2010a). Evidence of specificity of a visual P3 amplitude modulation deficit in
schizophrenia. Schizophrenia Research, 124, 119–126, http://dx.doi.org/10.1016/j.
schres.2010.08.014.
Neuhaus, A. H., Urbanek, C., Opgen-Rhein, C., Hahn, E., Ta, T. M. T., Koehler, S., et al.
(2010b). Event-related potentials associated with Attention Network Test.
International Journal of Psychophysiology, 76, 72–79, http://dx.doi.org/10.1016/
j.ijpsycho.2010.02.005.
Oberauer, K. (2005). Binding and inhibition in working memory: Individual and age
differences in short-term recognition. Journal of Experimental Psychology:
General, 134, 368–387.
Olesen, P. J., Westerberg, H., & Klingberg, T. (2004). Increased prefrontal and parietal
activity after training of working memory. Nature Neuroscience, 7, 75–79, http:
//dx.doi.org/10.1038/nn1165.
Pascual-Marqui, R. D. (2002). Standardized low-resolution brain electromagnetic
tomography (sLORETA): Technical details. Methods and Findings in Experimental
and Clinical Pharmacology, 24, 5–12.
Perrig, W. J., Hollenstein, M., & Oelhafen, S. (2009). Can we improve fluid
intelligence with training on working memory in persons with intellectual
disabilities? Journal of Cognitive Education and Psychology, 8, 148–164, http://dx.
doi.org/10.1891/1945-8959.8.2.148.
Persson, J., & Reuter-Lorenz, P. A. (2008). Gaining control: Training executive
function and far transfer of the ability to resolve interference. Psychological
Science, 19, 881–888.
Raven, J. C. (1990). Advanced progressive matrices. Sets I, II. Oxford: Oxford University
Press.
Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z.,
et al. (2012). No evidence of intelligence improvement after working memory
training: a randomized, placebo-controlled study. Journal of Experimental
Psychology: General, 142, 359–379, http://dx.doi.org/10.1037/a0029082.
Rueda, M. R., Fan, J., McCandliss, B. D., Halparin, J. D., Gruber, D. B., Lercari, L. P., et al.
(2004). Development of attentional networks in childhood. Neuropsychologia,
42, 1029–1040, http://dx.doi.org/10.1016/j.neuropsychologia.2003.12.012.
Rueda, M. R., Posner, M. I., Rothbart, M. K., & Davis-Stober, C. P. (2004). Development of the time course for processing conflict: an event-related potentials

2790

S. Oelhafen et al. / Neuropsychologia 51 (2013) 2781–2790

study with 4 year olds and adults. BMC Neuroscience, 5, 39, http://dx.doi.org/
10.1186/1471-2202-5-39.
Schmiedek, F., Li, S. C., & Lindenberger, U. (2009). Interference and facilitation in
spatial working memory: Age-associated differences in lure effects in the nback paradigm. Psychology and Aging, 24, 203–210, http://dx.doi.org/10.1037/
A0014685.
Schneiders, J. A., Opitz, B., Krick, C. M., & Mecklinger, A. (2011). Separating intramodal and across-modal training effects in visual working memory: An fMRI
investigation. Cerebral Cortex, 21, 2555–2564, http://dx.doi.org/10.1093/cercor/
bhr037.
Shipstead, Z., Hicks, K. L., & Engle, R. W. (2012). Cogmed Working Memory Training:
Does the Evidence Support the Claims? Journal of Applied Research in Memory
and Cognition, 1, 185–193, http://dx.doi.org/10.1016/j.jarmac.2012.06.003.
Shipstead, Z., Redick, T. S., & Engle, R. W. (2010). Does working memory training
generalize? Psychologica Belgica, 50, 245–276.
Shipstead, Z., Redick, T. S., & Engle, R. W. (2012). Is working memory training
effective? Psychological Bulletin, 138, 628–654, http://dx.doi.org/10.1037/
a0027473.22409508.
Silton, R. L., Heller, W., Towers, D. N., Engels, A. S., Spielberg, J. M., Edgar, J. C., et al.
(2010). The time course of activity in dorsolateral prefrontal cortex and

anterior cingulate cortex during top-down attentional control. Neuroimage,
50, 1292–1302, http://dx.doi.org/10.1016/j.neuroimage.2009.12.061.
Studer-Luethi, B., Jaeggi, S. M., Buschkuehl, M., & Perrig, W. J. (2012). Influence of
neuroticism and conscientiousness on working memory training outcome.
Personality and Individual Differences, 53, 44–49, http://dx.doi.org/10.1016/j.
paid.2012.02.012.
Szmalec, A., Verbruggen, F., Vandierendonck, A., & Kemps, E. (2011). Control of
interference during working memory updating. Journal of Experimental Psychology: Human Perception and Performance, 37, 137–151, http://dx.doi.org/10.1037/
a0020365.
Thorell, L. B., Lindqvist, S., Bergman, S., Bohlin, N. G., & Klingberg, T. (2009). Training
and transfer effects of executive functions in preschool children. Developmental
Science, 12, 106–113.
Unsworth, N., Redick, T. S., Heitz, R. P., Broadway, J. M., & Engle, R. W. (2009).
Complex working memory span tasks and higher-order cognition: A latentvariable analysis of the relationship between processing and storage. Memory,
17, 635–654, http://dx.doi.org/10.1080/09658210902998047.
Walsh, B. J., Buonocore, M. H., Carter, C. S., & Mangun, G. R. (2011). Integrating
conflict detection and attentional control mechanisms. Journal of Cognitive
Neuroscience, 23, 2211–2221.